[<prev] [next>] [day] [month] [year] [list]
Message-ID: <CAFkuX4uK6p8qzCgrJSUMY+6UaKw0dQN1PSC0zLrFD6N3ki7F9A@mail.gmail.com>
Date: Thu, 26 Jun 2014 12:55:58 -0600
From: "Don A. Bailey" <donb@...uritymouse.com>
To: oss-security@...ts.openwall.com
Subject: LMS-2014-06-16-4: FFmpeg LZO
Hello All,
A vulnerability has been identified in the FFmpeg LZO implementation.
Please find the bug report attached inline.
Best,
Don A. Bailey
Founder / CEO
Lab Mouse Security
https://www.securitymouse.com/
#############################################################################
#
# Lab Mouse Security Report
# LMS-2014-06-16-4
#
Report ID: LMS-2014-06-16-4
CVE ID: CVE-2014-4610
Researcher Name: Don A. Bailey
Researcher Organization: Lab Mouse Security
Researcher Email: donb at securitymouse.com
Researcher Website: www.securitymouse.com
Vulnerability Status: Patched
Vulnerability Embargo: Broken
Vulnerability Class: Integer Overflow
Vulnerability Effect: Memory Corruption
Vulnerability Impact: DoS, OOW, RCE
Vulnerability DoS Practicality: Practical
Vulnerability OOW Practicality: Practical
Vulnerability RCE Practicality: Practical
Vulnerability Criticality: Critical
Vulnerability Scope:
All versions of libav are affected.
All architectures supported by libav are affected.
Vulnerability Tested:
Yes. RCE proven on 10 separate platforms including but not limited to:
- Ubuntu and Mint x86, x86_64
- Debian x86_64, x86
- FreeBSD x86_64, x86
Functions Affected:
libavutil/lzo.c:av_lzo1x_decode
Criticality Reasoning
---------------------
This vulnerability can be triggered through a compression payload embedded
in a video file. Due to the nature of this memory corruption vulnerability,
exploitation of the bug can be seamless and work in the background during
normal video playback. A user will never notice that playback has been
compromised.
Vulnerability Description
-------------------------
An integer overflow can occur when processing any variant of a "literal run"
in the av_lzo1x_decode function. Each of these three locations is
subject to an integer overflow when processing zero bytes. The following code
depicts how the size of the literal array is generated:
static inline int get_len(LZOContext *c, int x, int mask)
{
int cnt = x & mask;
if (!cnt) {
while (!(x = get_byte(c)))
cnt += 255;
cnt += mask + x;
}
return cnt;
}
As long as a zero byte (0x00) is encountered, the variable 'cnt' will be
incremented by 255. Using approximately sixteen megabytes of zeros, 'cnt' will
accumulate to a maximum unsigned integer value in the 32bit variable.
Therefore, get_len() will return a negative 'cnt' value to its caller. The
checks in copy_backptr() will fail to properly test for negative 'cnt' values
resulting in the following test never catching an error:
if (cnt > c->out_end - dst) {
cnt = FFMAX(c->out_end - dst, 0);
c->error |= AV_LZO_OUTPUT_FULL;
}
av_memcpy_backptr does not check for negative 'cnt' values, which results in
a copy of one byte from 'src' to 'dst', evading a crash do to excessive
copying.
Finally, the copy function will never crash by calling memcpy with a negative
value because it only calls memcpy when the signed 'cnt' variable is greater
than zero. However, the pointers 'c->in' and 'c->out' will still be adjusted
by a negative value, causing 'c->out' to point to an area of memory prior to
the actual output buffer.
It is notable that since the count value 'cnt' is passed around as an 'int',
it will always be interpreted as a signed 32bit integer regardless of the
underlying architecture. This means that this vulnerability affects all
platforms and architectures regardless of whether they are 32bit or 64bit
in nature.
Vulnerability Resolution
------------------------
Resolving this issue requires several separate fixes.
1) lzo.c:get_len()
The return value of get_len must be evaluated for negative count values.
A negative value should never be allowed in this context. Always error
when a negative or zero value is returned.
2) lzo.c:copy()
A negative value should not be allowed as a parameter to copy(). In
addition, the pointers 'c->in' and 'c->out' should be tested after they
are changed by the count value. Verify that the new offset does not land
outside of the bounds of the 'out' buffer.
3) lzo:copy_backptr()
Do not allow a negative 'cnt' value to be passed to copy_backptr. Augment
the test cases to ensure that a negative value cannot be used to adjust
the 'c->out' pointer.
4) libavutil/mem.c:av_memcpy_backptr
Return an error value.
Do not allow a negative 'cnt' or 'back' value to be used.
5) Always use a size_t for any size variable.
Size variables should always represent the underlying architecture's largest
natural unsigned integer. Use size_t, or a variant, to automatically scale
the value to the underlying architecture.
Powered by blists - more mailing lists
Please check out the
Open Source Software Security Wiki, which is counterpart to this
mailing list.
Confused about mailing lists and their use?
Read about mailing lists on Wikipedia
and check out these
guidelines on proper formatting of your messages.
